8 ms·

This looks great, but haven't the big, embarrassing problems been with scaling the back end? Even if this were a superior UI and flow, it wouldn't address the i

by _pius 13y ago

This looks great, but haven't the big, embarrassing problems been with scaling the back end? Even if this were a superior UI and flow, it wouldn't address the issues that made the news.

hga 13y ago
Perhaps, and no. E.g. the front end registration problem is due to a late requirements change in August that had the effect of preventing window shopping, plus we know there were changes ordered though the week before launch, and no integration testing until 1-2 weeks before launch, and that discovered it locked up if 200 simultaneous login attempts were made or if it had ~1,100 users on it. So, yeah, those are scaling problems, but not ones we can necessairly blame the techies for, given that they didn't have enough time, and couldn't do integration testing (the government's CMS was the integrator and integration tester, now replaced by QSSI, the ones who did the "data hub"? I think it's called, which is supposed to be not so bad). But also no, in that the fix-it czar's #1 declared priority is to fix the garbage 834 EDI transactions that are being sent to insurers. That includes incorrect data and incorrect transactions (e.g. multiple enrollments and cancellations for the same person). Based on that, the front end's problems have been a blessing in disguise, because the insurers are only getting a few hundred enrollees a week and that low volume allows the insurers to call each up and straighten out problems. An estimated 16 million policies in the individual market, and an unknown number in Obamacare's interim high risk pool must sign up for new policies by December 15th or so or they'll suffer a lapse in coverage, which for many of the latter will be fatal. The only way to get subsidies is through the Federal system, i.e. the state exchanges use it for that (maybe with inaccuracies, but it's evidently working for the state exchanges that are working, but don't judge that for any given state exchange until you find out which are getting Obamacare and the much larger numbers reported to be signing up for Medicaid), and that makes sense to avoid fraud, since it results in direct payments from the Federal fisc.
- dr_ 13y ago
  Valid point with the window shopping comment. A huge mistake on the part of the administration though. The site was apparently flooded with visitors and you have to wonder how many of those visitors were people who already had decent coverage, privately or through their employer, but were just curious to check it out because they had heard so much about it. I know I was one of those people.
  - hga 13y ago
    Errr, as I read things, bypassing registration would have run up against the 1,100 simultaneous users limit ... rather hard. Of course that was presumably increased over time, it's not like they were able to do much performance tuning until those first end to end tests were conducted by CMS a week or two before launch.
    - grey-area 13y ago
      An 1100 user limit is absurd for a website this scale, and is likely predicated on serving content to logged in users, not anonymous readers. With some simple caching they could serve hundreds of thousands easily with very basic tech if they didn't require login. The biggest problem, from a usability and a technical perspective, is the hiding of the information till you have signed up and given details. Apparently this constraint was added at the last minute, I wonder if insurance companies insisted on it?
- jorgem 13y ago
  >> An estimated 16 million policies in the individual market What percentage of people with individual plans (who need to get new plans) would actually qualify for subsidies? I don't, so I have more options: (1) I can buy another individual plan from my insurer (2) I can shop other insurers or (3) I can buy from california exchange In none of those scenarios do I actually care whether healthcare.gov is working. I suspect I am not alone.
  - hga 13y ago
    True, I suspect the fraction is relatively smaller; when I put on my political hat, I call them kulacks. And they're more likely to be up to the task of finding all of this out on their own. But I think we can also assume the numbers who are not up to this, or are in the expiring high risk pool, are in the millions. People are going to suffer, because of lapses in insurance, or because they simply can't afford the costs of new Obamacare compliant policies (between the new essentials like coverage for your kids till age 26, and all the cost shifting), or are too proud, moral, or whatever to divorce to get their costs down. ADDED: This is going to really hit the early retirement thing and availability of slots for people to move up in. A very large fraction of the horror stories I'm reading are about early retirees who are not yet 65 and therefore moved to Medicare. Is CoveredCA really working that well? From an anti-Obamacare site, some statistics and analysis: http://insureblog.blogspot.com/2013/11/coveredca-statistics-along-with-some.html http://insureblog.blogspot.com/2013/11/coveredca-statistics-...
  - tanzam75 13y ago
    > What percentage of people with individual plans (who need to get new plans) would actually qualify for subsidies? I don't, so I have more options: (1) I can buy another individual plan from my insurer (2) I can shop other insurers or (3) I can buy from california exchange I don't understand why you're required to use the exchange to get the subsidies, anyway. Why not let people claim the subsidy directly through an insurer? Then turn healthcare.gov into a static site that simply displays the options and redirects to an insurer's web site. After all, the subsidies are only an estimate. The actual amount is only calculated on the 1040 tax return. So why should this be any different from tax withholding? We don't make everyone pay estimated taxes through a centralized site. We let employers withhold an estimated amount, and then the IRS balances the accounts annually, in a batch process.
Tloewald 13y ago
It doesn't need a back end at all. They've done exactly what I suggested in my blog but was too lazy to do — make a rough guess based on static data and hand you off to the vendor. It can be done client-side off static content. Using the IRS service to verify income would be something the real site could easily do, but it's not really necessary. http://loewald.com/blog/?p=5349 http://loewald.com/blog/?p=5349

by _pius 13y ago

This looks great, but haven't the big, embarrassing problems been with scaling the back end? Even if this were a superior UI and flow, it wouldn't address the issues that made the news.

hga 13y ago

Perhaps, and no. E.g. the front end registration problem is due to a late requirements change in August that had the effect of preventing window shopping, plus we know there were changes ordered though the week before launch, and no integration testing until 1-2 weeks before launch, and that discovered it locked up if 200 simultaneous login attempts were made or if it had ~1,100 users on it. So, yeah, those are scaling problems, but not ones we can necessairly blame the techies for, given that they didn't have enough time, and couldn't do integration testing (the government's CMS was the integrator and integration tester, now replaced by QSSI, the ones who did the "data hub"? I think it's called, which is supposed to be not so bad). But also no, in that the fix-it czar's #1 declared priority is to fix the garbage 834 EDI transactions that are being sent to insurers. That includes incorrect data and incorrect transactions (e.g. multiple enrollments and cancellations for the same person). Based on that, the front end's problems have been a blessing in disguise, because the insurers are only getting a few hundred enrollees a week and that low volume allows the insurers to call each up and straighten out problems. An estimated 16 million policies in the individual market, and an unknown number in Obamacare's interim high risk pool must sign up for new policies by December 15th or so or they'll suffer a lapse in coverage, which for many of the latter will be fatal. The only way to get subsidies is through the Federal system, i.e. the state exchanges use it for that (maybe with inaccuracies, but it's evidently working for the state exchanges that are working, but don't judge that for any given state exchange until you find out which are getting Obamacare and the much larger numbers reported to be signing up for Medicaid), and that makes sense to avoid fraud, since it results in direct payments from the Federal fisc.

dr_ 13y ago
Valid point with the window shopping comment. A huge mistake on the part of the administration though. The site was apparently flooded with visitors and you have to wonder how many of those visitors were people who already had decent coverage, privately or through their employer, but were just curious to check it out because they had heard so much about it. I know I was one of those people.
- hga 13y ago
  Errr, as I read things, bypassing registration would have run up against the 1,100 simultaneous users limit ... rather hard. Of course that was presumably increased over time, it's not like they were able to do much performance tuning until those first end to end tests were conducted by CMS a week or two before launch.
  - grey-area 13y ago
    An 1100 user limit is absurd for a website this scale, and is likely predicated on serving content to logged in users, not anonymous readers. With some simple caching they could serve hundreds of thousands easily with very basic tech if they didn't require login. The biggest problem, from a usability and a technical perspective, is the hiding of the information till you have signed up and given details. Apparently this constraint was added at the last minute, I wonder if insurance companies insisted on it?
jorgem 13y ago
>> An estimated 16 million policies in the individual market What percentage of people with individual plans (who need to get new plans) would actually qualify for subsidies? I don't, so I have more options: (1) I can buy another individual plan from my insurer (2) I can shop other insurers or (3) I can buy from california exchange In none of those scenarios do I actually care whether healthcare.gov is working. I suspect I am not alone.
- hga 13y ago
  True, I suspect the fraction is relatively smaller; when I put on my political hat, I call them kulacks. And they're more likely to be up to the task of finding all of this out on their own. But I think we can also assume the numbers who are not up to this, or are in the expiring high risk pool, are in the millions. People are going to suffer, because of lapses in insurance, or because they simply can't afford the costs of new Obamacare compliant policies (between the new essentials like coverage for your kids till age 26, and all the cost shifting), or are too proud, moral, or whatever to divorce to get their costs down. ADDED: This is going to really hit the early retirement thing and availability of slots for people to move up in. A very large fraction of the horror stories I'm reading are about early retirees who are not yet 65 and therefore moved to Medicare. Is CoveredCA really working that well? From an anti-Obamacare site, some statistics and analysis: http://insureblog.blogspot.com/2013/11/coveredca-statistics-along-with-some.html http://insureblog.blogspot.com/2013/11/coveredca-statistics-...
- tanzam75 13y ago
  > What percentage of people with individual plans (who need to get new plans) would actually qualify for subsidies? I don't, so I have more options: (1) I can buy another individual plan from my insurer (2) I can shop other insurers or (3) I can buy from california exchange I don't understand why you're required to use the exchange to get the subsidies, anyway. Why not let people claim the subsidy directly through an insurer? Then turn healthcare.gov into a static site that simply displays the options and redirects to an insurer's web site. After all, the subsidies are only an estimate. The actual amount is only calculated on the 1040 tax return. So why should this be any different from tax withholding? We don't make everyone pay estimated taxes through a centralized site. We let employers withhold an estimated amount, and then the IRS balances the accounts annually, in a batch process.

Tloewald 13y ago

It doesn't need a back end at all. They've done exactly what I suggested in my blog but was too lazy to do — make a rough guess based on static data and hand you off to the vendor. It can be done client-side off static content. Using the IRS service to verify income would be something the real site could easily do, but it's not really necessary. http://loewald.com/blog/?p=5349 http://loewald.com/blog/?p=5349